DDoS Detection using CNN

A deep learning-based DDoS detection system built using Convolutional Neural Networks (CNN). This project performs data preprocessing, trains a CNN model, and evaluates it on network traffic data to identify potential DDoS attacks.

Features

• Preprocessing support for large datasets (tested on 7GB+ CSV files using pandas chunking)
• Automatically handles varying column names and standardizes them using config.py
• CNN architecture with 2D convolution layers, pooling, and fully connected layers
• Modular scripts: preprocessing, training (model.py), and evaluation (testing.py)
• run.py script automates full pipeline

Getting Started

1. Clone the repository

git clone https://github.com/Avadhoot4757/DDOS-Detection-With-CNN.git
cd DDOS-Detection-With-CNN

2. Create a virtual environment (optional but recommended)

python3 -m venv myenv
source myenv/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Add your dataset

Place your large dataset (CSV file). We have included a balanced sample dataset sample_dataset.csv (under 10MB) containing equal entries from benign and DDoS classes, useful for quick testing under data/raw directory in root.

5. Run the project pipeline

python3 run.py

This will:

• Preprocess the dataset
• Train the CNN model
• Save and evaluate the model
• Start the testing interface

Resources

• Balanced Dataset: https://www.kaggle.com/datasets/devendra416/ddos-datasets
• Research Paper (Referenced): A_Deep_CNN_Ensemble_Framework_for_Efficient_DDoS_Attack_Detection_in_Software_Defined_Networks.pdf